Zero-shot denoising via neural compression: Theoretical and algorithmic framework

2025-06-15

AI Summary

Addressing the zero-shot denoising problem in medical and biological imaging—where neither training data nor clean reference images are available—this paper proposes ZS-NCD, a framework based on an untrained neural compression network. It performs block-wise online optimization and overlapping aggregation directly on a single noisy image. Innovatively, it leverages entropy constraints inherent in the compression model as an implicit regularization mechanism, eliminating the need for handcrafted regularizers or early-stopping heuristics. Furthermore, it establishes, for the first time, a finite-sample reconstruction error bound for compression-based denoisers. Extensive experiments demonstrate state-of-the-art zero-shot denoising performance under both Gaussian and Poisson noise, with strong generalization across natural and non-natural images—including biomedical modalities. The implementation is publicly available.

Technology Category

• Computer Vision: Medical and Biological Imaging
• Machine Learning: Learning on the Edge & Model Compression
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Zero-shot denoising aims to denoise observations without access to training samples or clean reference images. This setting is particularly relevant in practical imaging scenarios involving specialized domains such as medical imaging or biology. In this work, we propose the Zero-Shot Neural Compression Denoiser (ZS-NCD), a novel denoising framework based on neural compression. ZS-NCD treats a neural compression network as an untrained model, optimized directly on patches extracted from a single noisy image. The final reconstruction is then obtained by aggregating the outputs of the trained model over overlapping patches. Thanks to the built-in entropy constraints of compression architectures, our method naturally avoids overfitting and does not require manual regularization or early stopping. Through extensive experiments, we show that ZS-NCD achieves state-of-the-art performance among zero-shot denoisers for both Gaussian and Poisson noise, and generalizes well to both natural and non-natural images. Additionally, we provide new finite-sample theoretical results that characterize upper bounds on the achievable reconstruction error of general maximum-likelihood compression-based denoisers. These results further establish the theoretical foundations of compression-based denoising. Our code is available at: github.com/Computational-Imaging-RU/ZS-NCDenoiser.

Problem

• Denoise images without training data or clean references
• Propose neural compression-based framework for zero-shot denoising
• Provide theoretical bounds for compression-based denoising performance

Innovation

• Untrained neural compression network optimization
• Built-in entropy constraints prevent overfitting
• State-of-the-art zero-shot denoising performance